Medicine is powered by knowledge, but how do we know what is true and what is not? How do we deal with uncertainty in a setting where outcomes are not closely related to known variables? For example, although there are a few people who have survived jumping or falling from an airplane at high altitude (http://zidbits.com/2010/12/can-you-survive-a-freefall-without-a-parachute/), it is a rare event. Thus, a test to determine how to prevent death from such a disaster would only take a small number of participants to see if a particular method works. In contrast, when considering a medical condition where a large fraction of people might seemingly "recover" without treatment, such as tuberculosis (http://www.who.int/mediacentre/factsheets/who104/en/print.html), how does one determine if a treatment is effective? In this talk, I will examine how we gained knowledge about tuberculosis as an example of a disease where a combination of observational scientific findings and clinical trial data are linked to advance knowledge. I will also discuss other examples of clinical trials challenges and the solutions to these challenges.